Waste in Swiss Forest is a concern for wildlife and the environment, but also for the perceived beauty of nature. Litter such as plastic bottles, cans or food packaging is often left behind by people going through the forest, and is detrimental to the ecosystem. This project aims to portray the state of waste in Zurichs forests at Hoenggerberg and Kaeferberg in June 2025, and to find solutions for the problem. For this, people in Zurich were asked about how they perceive the state of waste in the forests of Hoenggerberg and Kaeferberg. It was investigated, where they saw which types of waste in the forest and what activities were likely causing this waste. A special focus was placed on the impact of sports in the forest. Finally, participants selected possible measures to remove waste from the forest.
It was found that most litter in the forest lands around picnic areas and benches, in the form of consumer goods such as plastic bottles, cans and cigarette butts. While participants generally have a positive view of the waste situation in the forest (4 out of 5), they believe that activities such as barbecue, picnic or camping cause the most waste for the forest. Sports activities were not perceived to have an extra impact on waste.
For measures against waste, participants prefer actions that are tied to personal responsibility first, such as trash-removal by the ones who left it, and to install more bins in the forest for this purpose. Fines for those who litter should further enforce this direct responsibility. Clean-up of waste by authorities or volunteers received the second-most votes.
Methods
Data gathering
For this project, a total of 27 participants were asked about their assessment of waste in forests around ETH Hoenggerberg. 12 people were approached in the forest on the evening of Sunday, 1st of June 2025. They were asked to fill out a Google survey about how they perceive waste in that forest in general, and specifically on that day. The following screenshot displays the briefing that the participants received before starting the survey. The photo on the right shows me in the process of looking for participants in the forest.
Site plan
The following satellite view show the site of the survey: The survey was conducted in the two forests “Hoenggerberg forest” and Kaeferberg forest in the north of Zurich.
The survey was conducted near the Campus of ETH Hoenggerberg. The student housing on the campus is marked with a red square and is situated close to the Hoenggerberg forest. In fact, many participants of this survey are living in these apartment blocks.
The QR-codes indicate where flyers with a link to the online survey were hung up. The list icon with a green frame indicates the two positions near the forest entrances where participants were asked to take the survey in person.
The survey consists of 19 questions, separated into 5 sections:
Waste in the forest
(6 questions, 2 optional)
Forest activities
(4 questions)
Sports in the forest
(4 questions)
Measures against waste in the forest
(3 questions)
A section about the user’s demographic information
(2 questions: age, gender)
A second survey was created in German, in order to reach more participants. However, the workflow to entertain two surveys simultaneously was later deemed as too complicated. Hence, only two participants (an elderly couple) answered the German survey on the first day, and all following participants were shown the English survey instead.
The answers of the 2 German-speaking participants were manually entered into the English version of the Google Form on Thursday, 6th of June, by the study author. Thus, all analysis could be done based on the results from the English survey.
On Monday, 2nd of June 2025, the English survey was then also sent to residents in student housing at ETH Hoenggerberg via social media. Further, flyers with a QR code were hung up in the student housing complex. More flyers were also hung up at different picnic-area spots in the Hoenggerberg and Kaeferberg forests.
To ensure validity, participants who filled out the survey in this manner were reminded again to only fill out the survey if they were in the forest on that day. But it since they were not met in the forest in person, like the first batch of participants, it can’t be verified if they actually spent time in the forest on the day they filled out the survey.
Data processing
The survey was then processed with the help of R-scripts on the R-Studio platform. The first R-script “01-data-download.R” extracts the raw data from the Google Sheet that comes along with the survey. The second R-script “02-data-cleaning” then processes this data into a tidy dataframe, one cell per value, one value per cell. Finally, a Quarto document “index.qmd” is used to write this report within R-Studio.
The R-packages used for this process were tidyverse, readr, ggplot2, gt, knitr, googlesheets4, dplyr and lubridate. During development, Git version control was applied and the progress was stored on a GitHub repository.
The final report was then published via GitHub pages. The R-code and the generated visualizations, with their interpretation, are listed below.
R-packages
library(tidyverse)
── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr 1.1.4 ✔ readr 2.1.5
✔ forcats 1.0.0 ✔ stringr 1.5.1
✔ ggplot2 3.5.1 ✔ tibble 3.2.1
✔ lubridate 1.9.4 ✔ tidyr 1.3.1
✔ purrr 1.0.4
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag() masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
In this section, visualization tools learned in the RBTL course are utilized to create different tables and diagrams to highlight certain trends in the data.
Demography and timeline
The following point plot Figure 1 displays when survey was taken and how the users rated the waste situation in the forest around Hoenggerberg. The rating range was from 1 to 5, with 5 being the best (highest cleanliness). It is likely that the given rating reflected the tidyness at in the forest at that time. Most people rated the cleanliness of the Hoenggerberg and Kaeferberg forest between 3 and 4, so there is room for improvement. But the lowest rating of 1 was not given a single time. No trends are visible regarding the gender of the participants.
ggplot(data = survey_small, mapping =aes(x = timestamp,y = waste_rating,color = gender,shape = gender ) ) +geom_point(size =3) +labs(x ="survey timestamp", y ="perceived forest cleanliness") +scale_shape(solid =FALSE) +ylim(1, NA)# the tip to include scale_shape(solid = FALSE) came from stackoverflow,# https://stackoverflow.com/a/51775750
Figure 1: Survey timeline and perceived cleanliness of the forest
The figure is a bit hard to read, because most people filled the survey at around the same time. To make it more clear, the following table Table 1 presents the summarized data in a more understandable form:
All participants were either male or female. The survey also presented the options to chose “non-binary” or “prefer not to answer” for the gender, but these options were never selected.
The second figure Figure 2 also displays the number of answers per day, while also showing a more visible distinction between male and female responses. 12 people were met in-person in the forest on Sunday evening, of which most were male. One additional person answered the survey online later. From Monday to Thursday, responses came only from people answering the questionnaire online. Most of these answers came in on Monday, when the link to the survey was posted in a WhatsApp group chat for students living in student housing on Campus Hoenggerberg, which is only 200 meters away from the Kaeferberg forest. Most of these participants were also male.
# same dataframe as in the code block above, so I don't use write_csvsurvey_weekday <- survey_small |>group_by(weekday, gender) |>summarize(amount =n())
`summarise()` has grouped output by 'weekday'. You can override using the
`.groups` argument.
# adding the white labelling in this plot was done with the help of ChatGPT, prompt:# https://chatgpt.com/share/6841b0b5-c2e4-8011-b70d-c15aacb89456ggplot(data = survey_weekday, mapping =aes(x = weekday,y = amount,fill = gender)) +geom_col(position ="stack") +geom_text(aes(label = amount), position =position_stack(vjust =0.5), color ="white") +labs(title ="Number of answers per day",x ="weekday the survey was taken",y ="number of answers")
Figure 2: number of answers per day
The following histogram Figure 3 highlights the demography of the participants:
survey_demography <- survey_smallwrite_csv(survey_demography, here::here("data/final/fig-demography.csv"))ggplot(data = survey_demography,mapping =aes(x = age, fill = gender)) +geom_histogram(color ="black") +labs(title ="age of participants in years", subtitle ="27 participants total",x ="age in years", y ="count") +theme_minimal()
`stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
Figure 3: age of participants in years
Most participants were between the ages of 20-40, since a lot of participants came from the student housing apartments near the forest entrance.
Statistical summary
The following tables and diagrams provide a statistical summary of the numerical values collected from the survey.
In the first question of the survey, people were shown a picture of a plastic cup on forest ground. They were asked about their feeling when seeing trash in the forest, from 1 to 5 (1 = very upset, 5 = very happy.)
Figure 4 shows that all participants reported a bad feeling (level 1 or 2) after seeing waste in the forest. There are no obvious trends from gender. But it is remarkable that no person above 40 reported a feeling level of 2. It could be concluded that the older the forest user, the more they are appaled by seeing waste in the forest. However, the small number of participants cannot justify this claim.
ggplot(data = survey_small, mapping =aes(x = age,y = waste_feeling,color = gender,shape = gender ) ) +geom_point(size =3) +labs(title ="Particpant's mood level when seeing waste in the forest",x ="age in years", y ="feeling when seeing forest waste")
Figure 4: Relation between age, gender and feeling when seeing waste in the forest
The first table Table 2 lists the average and median ages of the participants, grouped by gender. The average age is close to the median age, around 33 to 34. However, the standard deviation is quite large, so there are outliers: Some participants were much older than others. The maximum age was 63, as shown above in @fig_demography
survey_table_gender1 |>gt() |>fmt_number(decimals =1) |>tab_header(title ="statistical table of participants age, their forest-rating and their feeling when seeing forest waste",subtitle ="data from 27 forest users in June 2025, grouped by gender")
Table 2: Statistical data of the survey part 1
statistical table of participants age, their forest-rating and their feeling when seeing forest waste
data from 27 forest users in June 2025, grouped by gender
gender
count
age_mean
age_sd
age_median
rating_mean
rating_sd
rating_median
feeling_mean
feeling_sd
feeling_median
female
9.0
33.3
13.6
34.0
3.6
0.9
4.0
1.2
0.4
1.0
male
18.0
26.4
10.5
22.0
3.7
0.9
4.0
1.4
0.5
1.0
The table also provides info about how the participants rate the cleanliness of Zurichs forests on average, and the median. Since the standard deviation is low and the mean is close to the average, it can be concluded that most people rate the cleanliness between 3.5 and 4 and there are no significant outliers. However, the range (from 1 to 5) is quite small, so this has to be kept in mind.
The greatest agreement is in the participant’s feeling when seeing waste in the forest: All participants report a bad feeling (1 or 2) when seeing forest waste. The average and the median are close to 1, and the standard deviation is low.
In conclusion: While participants deem that Zurichs forests are clean, any small amount of waste seen is perceived as very negative. So, according to participants, even small amounts of waste have to be avoided at all costs.
The same conclusions can be made from the following graphic, which shows yet again how participants rate the cleanliness of Zurichs forests, charted by age and gender.
The next table Table 3 shows more statistical insights. It shows that the numbers for weekly forest visits and the value of forest sports are close for both women and men. Men tend to visit the forest slightly more often than women, and rate the value of sports activities in the forest higher. But the difference between the two genders is small when compared to the standard deviations.
A bigger difference can be seen for the suggested number of removal-days per year. Women are generally in favor of more waste-removal-days per year than the men, as shown in the mean and median values. However, the male participants propose a larger range of removal days by authorities, with a minimum of 6 and a maximum of 180. Although, since there are many more men that participated in the survey, this bigger number of responses could be the cause for the wider numerical span, and doesn’t necessarily have to be due to the gender.
survey_table_gender2 |>gt() |>fmt_number(decimals =1) |>tab_header(title ="Participants visiting frequency, their value of forest sports, and their suggested waste-removal days per year",subtitle ="Data from 27 forest users in June 2025, grouped by gender")
Table 3: Statistical data of the survey, part 2
Participants visiting frequency, their value of forest sports, and their suggested waste-removal days per year
Data from 27 forest users in June 2025, grouped by gender
gender
count
visits_mean
visits_sd
visits_median
sports_value_mean
sports_value_sd
sports_value_median
removal_days_mean
removal_days_median
removal_days_min
removal_days_max
female
9.0
2.8
1.2
3.0
4.1
0.9
4.0
46.9
52.0
12.0
100.0
male
18.0
3.1
2.1
2.5
4.1
1.3
5.0
37.6
24.0
6.0
180.0
The distribution of suggested removal days is also shown in the following diagram Figure 6 . It seems that most people desire regular waste-removal by authorities every week, indicated by the large amount of data points around the number 52. But some are also satisfied with a monthly waste-removal schedule, which corresponds to 12 removal days per year.
ggplot(data = survey_small, mapping =aes(x = age,y = measures_frequency,color = gender,shape = gender ) ) +geom_point(size =3) +labs(title ="Suggested number of days per year for waste-removal by authorities",x ="age in years", y ="suggested days of waste-removal")
Figure 6: Suggested number of days per year for waste-removal by authorities
The next plot Figure 7 shows the relation between the participants’ forest-visiting frequency and their rating of the forest’s cleanliness. Jitter was used to highlight stacked data points, where multiple participants have the same days in the forest and the give the same cleanliness rating. Generally, people who visit the forest more often tend to give a higher rating. One interpretation of this would be that people who visit the forest more often can better assess the situation. People who visit the forest less often may have a distorted view of litter in the forest, with a trend to a more negative rating.
ggplot(data = survey_small, mapping =aes(x = activities_frequency,y = waste_rating, ) ) +geom_point(shape =1, size =3) +geom_jitter(width =0.1, height =0.1) +labs(title ="Forest cleanliness, depending on visiting frequency",subtitle ="answers from 27 forest users",x ="Days per week spent in the forest",y ="Perceived forest cleanliness")
Figure 7: forest cleanliness rating by groups with different visiting frequencies
Waste origins
In this section, the reported waste types and locations are analyzed.
The following Figure 8 shows the different trash items that forest users found on that day. Not much more can be interpreted from this data, other than there seems to be a trend to consumer items such as plastic bottles, cigarettes, cans and food packaging landing in the forest more often. However, almost nobody used the “other” option in the survey to manually report other trash items found in the forest. In fact, only 1 person used this option, and wrote “horse shit”.
ggplot(data = waste_types,mapping =aes(x = waste_type_today,y = count)) +geom_col() +labs(x ="Waste type seen today")
The following graph Figure 9 displays how forest users report where they see waste in the forest in general, and on the day they took the survey. What stands out is that people think that waste is concentrated mainly around picnic areas. But on the other hand, on the day they took the survey, they saw most of the waste along paths or trails. There might be some bias, coming from past experiences where people saw very dirty picnic areas in the forest.
# I used ChatGPT to combine the 2 dataframes, prompt:# https://chatgpt.com/share/68421141-9c14-8011-9448-3ecde5165adawaste_location$source <-"forest waste in general"waste_location_today$source <-"forest waste today"waste_combined <-rbind(waste_location, waste_location_today)ggplot(data = waste_combined, aes(x = waste_location, y = count, fill = source)) +geom_bar(stat ="identity", position ="dodge") +labs(title ="Comparison of waste locations in the forest, in general and today",x ="location", y ="count") +theme_minimal()
Figure 9: Comparison of waste locations in the forest, in general and today
Conclusion
As a result of this survey, it was found that most people think that litter in the forest is concentrated around picnic spots and benches, and mainly consists of consumer waste like plastic bottles, cans, and cigarette butts. Although the participants generally rated the forest’s waste situation in a positive way (with an average score of 4 out of 5), they report deeply negative feelings (scoring 1 or 2 out of 5) when seeing waste in the forest. The waste was identified to come from activities such as barbecuing, picnicking, and camping as the primary sources. In contrast, sports activities were not seen as a major contributor to the problem.
When it comes to removing waste, participants favored solutions that emphasize individual responsibility. They supported that people clean up their own litter, and further suggested adding more waste bins throughout the forest. Fines weree seen as a good measure to enforce this accountability. The second most preferred option were cleaning activities carried out by authorities or volunteers.
The demographical data showed no significant trends towards gender. Older people tended to react more negatively towards waste, and gave the forests a lower cleanliness rating. But since the sample size was small, the significance assessment should not be overexaggerated.